I found many posts explaining what to do with replicates, but none of them explored this approach. Maybe I missed it.
I want to pseudo-pool 3 technical PCR replicates in DADA2 and only keep the ASV which were found in 2 out of 3 replicates. Is that possible with QIIME2?
1 Like
Hi @Adrian1,
Interesting approach! The short answer is sort of, but I think it becomes a question of ...and then what.
I think if I were doing this, I'd split your table based on the sample. (So, if you have samples A, B, and C with replicates A1, A2, and A3, B1, B2, B3, and so on, I'd make a table taht's A1, A2, and A3 and then another thats B1, B2, and B3, and so on). You can do that using filter-samples in the q2-feature-table plugin. I'd think about scripting it if I were you, but YMMV.
Once you decide what "present" means, I'd either use filter-features (present=1 read) or filter-features-conditionally (present = relative abundance, i.e. feature needs to be present at 1/1000 whether there are 1001 sequences or 10,0001 sequences in the replicate.)
You can then decide if you want to pool there, or merge the tables back together and then use the group function to merge the replicates.
I'd be super intrested to see how this affects the results vs just a global filtering of low abundance data and then maybe using a random intercept model to deal with your replicates.
Best,
Justine
5 Likes